QSPR Cholesterol Esterase Inhibition Constant Calculation for Dental Monomer Biocompatibility

Dental biocompatibility tests designed to minimize membrane component damage are needed as membranes are the first line of cell defense. Research to design non-inhibiting resin monomers of key membrane bound enzymes such as cholesterol esterase (CHase) is one critical aspect of this testing. Objective: Develop a Quantum mechanical Quantitative Structure Property Relationship (QSPR) computational model to predict inhibition constants (Ki) for chemical interactions with Chase. This QSPR should be statistically sound and extensible to dental component structures so that ultimately it can be used as a model to design those which are not inhibitors of human membrane CHase. Hypothesis tested: Reflecting the experimental error in literature Ki's a predictive QSPR for Ki(R² > 0.700) is feasible. Methods: For the QSPR the literature Ki's of 60 chemical structures for porcine CHase (functionally similar to human Chase) were selected. The AM1 semiempirical quantum mechanical method was used to optimize all literature chemical structures. The QSPR program CODESSA was applied to develop a regression equation linking chemical descriptors with literature Ki values. Results: The sixty chemical structures and associated literature Ki ‘s yielded a QSPR model with the following characteristics: R² = 0.831, adjusted R² = 0.820, F ratio = 73.8. Two chemically meaningful descriptors, their significance (p) and variance inflation factors (VIF) were: 1.) Max valency for an oxygen atom (p<0.0001, VIF = 1.07); 2.) XYshadow/XYrectangle (p<0.0001, VIF = 1.07). Use of the QSPR led to correct prediction of the Ki for the monomer BisPhenol A Dimethacrylate (BADM). Conclusions: AM1 Ki calculations are feasible as hypothesized. The chemical descriptors identified are appropriate for cholesterol inhibition constant predictions. They involve O bond stability and molecular size/shape. Design of biocompatible molecules which do not inhibit human membrane cholesterol esterase is feasible using the QSPR Ki model. Supported by NIH/NIDCR Grant RO1 DEO 14379.